SQLBot

We'd reach for SQLBot when the data is already in MySQL or PostgreSQL and the blocker is people, not pipelines — the analyst queue that fills up with "can you pull last week's channel numbers" requests. Self-hosting is the other half of the case: nothing leaves your infrastructure, and workspace isolation, fine-grained permissions and audit logs give a governance story you can take to a security review. The part that earns its keep is the knowledge base. Feeding it business terms, table schemas, custom prompts and example SQL is unglamorous work, but it's what turns plausible-sounding SQL into correct SQL on your actual tables. SQLBot ships two built-in connectors, MySQL and PostgreSQL, so a warehouse on anything else is a pass unless you can bridge it. Where it bites: charting is basic. There's a dashboard builder that lays out charts from several conversations, and for many internal questions that's enough, but teams expecting Tableau-grade output will be disappointed. Pair it with DataEase — same vendor, same design language, and the docs already treat it as a companion. Setup is the other caveat. This is self-hosted software with a model to configure and a knowledge base to maintain; the more you put in, the better the answers get. If you want zero-configuration conversational BI, this isn't that, and the maintenance never fully stops. The closest alternative depends on why you're looking. DataEase if you want the wider Fit2Cloud BI suite, MaxKB if you're building an agent platform rather than a query tool. Both come from the same team, which makes mixing them low-friction. Use it for multi-turn exploration of a MySQL or Postgres dataset — ask, then follow up on the result with analysis, explanation and verification. Skip it on latency-sensitive, real-time